Energy-preserving bijection between RSOS paths and transfer-matrix string patterns in the physical combinatorics of critical RSOS and XXX models

Feverati, Pearce and Witte ('Physical Combinatorics and Quasiparticles', J. Stat. Mech. 2009) unified the corner-transfer-matrix and Bethe-ansatz descriptions of the spectra of critical lattice models through a conjectured energy-preserving bijection between one-dimensional RSOS configuration-sum paths and the patterns of zeros of fused transfer-matrix eigenvalues in their analyticity strips, verified numerically only for sizes N <= 16 on A-type and tadpole T-type graphs. The same paper derives the vacuum-sector conformal energies of the unitary minimal models and of the XXX spin chain from the thermodynamic Bethe ansatz (TBA), but leaves open the identification of Bethe-root j-strings with transfer-matrix 2-strings for all N, the existence of the large-q stabilization limits of the infinite-fusion TBA of the XXX model, and the (m,n)-systems, integration constants and energy formulas for arbitrary conformal sectors (r,s). The problem is the rigorous completion of this classification program, and the general conjectures remain open.

Background

Exactly solvable two-dimensional lattice models at criticality and their continuum conformal field theories (CFTs) admit two parallel exact bookkeepings that were historically developed separately.

The first is Baxter's corner transfer matrix (CTM) approach - a transfer matrix built on a 90-degree wedge of the lattice, from which the finite-size partition function is reassembled. For the Andrews-Baxter-Forrester (ABF) restricted-solid-on-solid (RSOS) models - lattice models whose local variable, the height, lives on the nodes of a Dynkin diagram and may change only between adjacent nodes - the CTM reduces the partition function in regime III (the antiferromagnetic critical regime) to a sum over one-dimensional lattice paths weighted by a local energy statistic. These one-dimensional configurational sums are finitized characters: finite-size truncations of CFT characters whose $N \to \infty$ limits admit fermionic (positive-term) expressions interpreted as quasiparticle partition functions, in the sense introduced by Dasmahapatra, Kedem, Klassen, McCoy and Melzer. For the RSOS model on the diagram $A_L$ (a line of $L$ nodes), the scaling limit is the unitary minimal CFT $\mathcal{M}(L,L+1)$ of central charge $c = 1 - 6/(L(L+1))$; for the level-1 $\widehat{su}(2)$ Wess-Zumino-Witten (WZW) model - a conformal field theory whose chiral symmetry algebra is the affine (loop) extension of $su(2)$ - with $c = 1$, the lattice regularization is a six-vertex model with rational Boltzmann weights that reduces in the Hamiltonian limit to the spin-1/2 XXX quantum spin chain (the isotropic Heisenberg chain), whose path bookkeeping uses the tadpole diagram $T_L$ (a chain of $L$ nodes with a single loop at one end).

The second is the Bethe ansatz. Eigenvalues of the commuting family of fused transfer matrices (transfer matrices built from the fundamental one by the fusion procedure, the algebraic construction of the higher-spin conserved quantities of the model) are classified by patterns of zeros of the eigenvalues in analyticity strips of the complex spectral parameter $u$: 2-strings (pairs of bound-state zeros) behave as particles, 1-strings as dual-particles, and the allowed particle content of each strip is fixed by an (m,n)-system - a linear Diophantine system built on the graph's Cartan matrix (the integer matrix encoding the node adjacencies of the diagram), encoding nearest-neighbour exclusion and packing rules. For the XXX chain, the eigenvalues of Baxter's auxiliary matrix $Q(u)$ satisfy the TQ relation with the transfer matrix $T(u)$ - the functional equation linking $T(u)$ and $Q(u)$, from which the Bethe equations for the zeros of $Q(u)$ follow - and its zeros are the Bethe roots, which under the string hypothesis organize into strings: clusters of zeros sharing a common real part, a cluster of $j$ zeros being a $j$-string. This analyticity information is the input to the thermodynamic Bethe ansatz (TBA), a system of nonlinear integral equations - one for the pseudo-energy $\varepsilon^q(x)$ of the particles in strip $q$ - whose solution yields the exact conformal energies.

At the level of counting the two bookkeepings agree: polynomial boson-fermionic character identities - Melzer's polynomial identities, proven by Warnaar for the vacuum sector, with the general-sector fermionic counting of RSOS states due to Berkovich - establish that the finitized characters computed from paths equal those computed from string patterns. However, these are analytic identities between generating functions; they do not supply an explicit structural map between individual paths and individual string patterns. Dasmahapatra and Foda established such a weight-preserving bijection explicitly (standard tableaux - standard fillings of Young diagrams - interpolating between paths and rigged configurations, i.e. strings equipped with integer quantum numbers) but only for the vacuum sectors of regime-III ABF models, conjecturing in 1996 that analogous bijections exist for the remaining sectors and more general models. Feverati and Pearce exhibited the bijection in 2003 only for vacuum-sector Ising-type examples. Feverati, Pearce and Witte (2009) then gave the general construction: an algorithm decomposes any RSOS path on $A$- and $T$-type graphs into particles, dual-particles and tower particles, and the resulting configuration is identified with a string pattern through an energy-preserving bijection - stated, however, as a conjecture, checked only numerically up to $N = 16$. Using the string patterns alone (the TBA derivation is logically independent of the bijection), they solved the vacuum-sector TBA completely and derived the general conformal energy expressions of the minimal RSOS models and of the XXX model. For the XXX model the fusion hierarchy does not truncate, and their derivation rests on the additional assumption that the zero positions $y_h^{(q)}$ and pseudo-energies $\varepsilon^q(x)$ in the strips stabilize as the strip index $q \to \infty$, which justifies a tadpole truncation of the infinite TBA system. They also observed and conjectured that the $j$-strings of Bethe roots of $Q(u)$ coincide with the transfer-matrix 2-strings in strip $j$.

The open problem is the rigorous completion of this program: the bijection itself, its companion XXX conjectures, and the extension of the classification from the vacuum sector $(r,s) = (1,1)$ to arbitrary conformal sectors $(r,s)$ (the sectors labelled by the Kac indices $r,s$ of the conformal dimensions $h_{r,s}$).

Problem Statement

Establish rigorously - or refute - the conjectured identifications that underpin the quasiparticle classification of the spectra of the critical ABF RSOS models and the XXX spin chain, and complete that classification beyond the vacuum sector. In the formulation of the source (Feverati-Pearce-Witte, arXiv:0903.0510v2, Sections 2.5, 2.11, 3.3.8, 4.6), four linked questions remain open:

1. Bijection. Prove that the particle-decomposition algorithm of the source paper defines, for every system size $N$ in the source's setup (even $N$, vacuum sector, fixed boundary conditions $a_0 = a_N = 1$, $a_{N+1} = 2$) and every $A$-type or tadpole $T$-type graph, a bijection between the set of one-dimensional RSOS configuration-sum paths and the set of distinct admissible string patterns - patterns of zeros of the fused transfer-matrix eigenvalues in their analyticity strips - and that this bijection preserves the energy statistic; or exhibit a counterexample, i.e. a graph, a size $N$, and explicit path data for which the algorithm fails injectivity, surjectivity, or energy preservation.

2. Bethe strings. Prove or refute that, for the XXX model, for all finite $N$, an eigenvalue of Baxter's auxiliary matrix $Q(u)$ has a $j$-string at the same relative position as each 2-string in strip $j$ of the transfer-matrix eigenvalues, and that in the limit $N \to \infty$ the locations of the $j$-strings of $Q(u)$ exactly coincide with the locations of the 2-strings of the transfer matrices in strip $j$.

3. Large-$q$ limits. Prove or refute the source's assumption (4.67): as the strip index $q \to \infty$, the zero positions $y_h^{(q)}$ and the pseudo-energies $\varepsilon^q(x)$ approach well-defined limits $y_h^{(\infty)}$ and $\varepsilon^\infty(x)$, as required to justify the tadpole truncation of the infinite TBA system of the XXX model.

4. All sectors. Derive the $(m,n)$-systems, the TBA integration constants, and the complete excited-state energy formulas for arbitrary conformal sectors $(r,s)$ of the minimal RSOS models on the strip; the results must be consistent with the known conformal dimensions

$$h_{r,s} = \frac{[(L+1)r - Ls]^2 - 1}{4L(L+1)}$$

of the unitary minimal models and with finite-$N$ transfer-matrix eigenvalue data.

The four questions are of equal standing as sub-questions and each is independently resolvable: resolving or refuting any one of them stands on its own, neither requiring the resolution of the others nor settling them. The connections drawn below are thematic, not logical dependencies, so a refutation of Question 1 leaves Questions 2-4 to be resolved on their own terms.

These four questions form one program. Question 1 is the load-bearing combinatorial foundation: the classification of eigenstates through paths, on which the physical interpretation of the quasiparticles rests. Question 2 is its analytic counterpart on the XXX side: it fixes the particle content from the zero content of $Q(u)$ and welds the Bethe ansatz to the combinatorial classification. Question 3 underpins the infinite-fusion TBA by which the XXX energies are evaluated. Question 4 extends the classification, which the source completes only for the vacuum sector $(r,s) = (1,1)$ with fixed boundaries, to the remaining sectors. The source additionally suggests extensions to other boundary conditions (periodic and toroidal) and to other model families ($D$- and $E$-type RSOS models, $Z_k$ parafermions, higher rank); these remain open directions but do not belong to the completion boundary of this problem.

References

1. G. Feverati, P. A. Pearce, N. S. Witte, Physical Combinatorics and Quasiparticles, J. Stat. Mech. (2009) P10013. DOI: 10.1088/1742-5468/2009/10/P10013; arXiv:0903.0510; https://arxiv.org/abs/0903.0510
2. S. Dasmahapatra, O. Foda, Strings, paths, and standard tableaux, Int. J. Mod. Phys. A13 (1998) 501-522. DOI: 10.1142/S0217751X98000214; arXiv:q-alg/9601011; https://arxiv.org/abs/q-alg/9601011
3. G. Feverati, P. A. Pearce, Critical RSOS and minimal models: fermionic paths, Virasoro algebra and fields, Nucl. Phys. B663 (2003) 409-442. DOI: 10.1016/S0550-3213(03)00375-4; arXiv:hep-th/0211185; https://arxiv.org/abs/hep-th/0211185
4. G. E. Andrews, R. J. Baxter, P. J. Forrester, Eight-vertex SOS model and generalized Rogers-Ramanujan-type identities, J. Stat. Phys. 35 (1984) 193-266. DOI: 10.1007/BF01014383; https://doi.org/10.1007/BF01014383
5. S. Dasmahapatra, R. Kedem, T. R. Klassen, B. M. McCoy, E. Melzer, Quasi-particles, conformal field theory, and q-series, Int. J. Mod. Phys. B7 (1993) 3617-3648. DOI: 10.1142/S0217979293003437; https://doi.org/10.1142/S0217979293003437
6. E. Melzer, Fermionic character sums and the corner transfer matrix, Int. J. Mod. Phys. A9 (1994) 1115-1136. DOI: 10.1142/S0217751X94000510; https://doi.org/10.1142/S0217751X94000510
7. S. O. Warnaar, Fermionic solution of the Andrews-Baxter-Forrester model. II. Proof of Melzer's polynomial identities, J. Stat. Phys. 84 (1996) 49-83. DOI: 10.1007/BF02179577; https://doi.org/10.1007/BF02179577
8. A. Berkovich, Fermionic counting of RSOS states and Virasoro character formulas for the unitary minimal series M(v, v+1): exact results, Nucl. Phys. B431 (1994) 315-348. DOI: 10.1016/0550-3213(94)90108-2; https://doi.org/10.1016/0550-3213(94)90108-2
9. Z. Bajnok, O. el Deeb, P. A. Pearce, Finite-Volume Spectra of the Lee-Yang Model, JHEP 04 (2015) 073. DOI: 10.1007/JHEP04(2015)073; arXiv:1412.8494; https://arxiv.org/abs/1412.8494
10. O. el Deeb, The critical boundary RSOS M(3,5) model, Theor. Math. Phys. 193 (2017) 1811-1825. DOI: 10.1134/S0040577917120078; arXiv:1512.02185; https://arxiv.org/abs/1512.02185
11. G. Z. Fehér, P. A. Pearce, A. Vittorini-Orgeas, One-dimensional sums and finitized characters of 2 x 2 fused RSOS models, J. Stat. Mech. (2018) 073104. DOI: 10.1088/1742-5468/aace1d; arXiv:1804.03332; https://arxiv.org/abs/1804.03332
12. J. Lamy-Poirier, P. Mathieu, Path representation of su(2)_k states I: Operators and particles for k = 1, 2, Nucl. Phys. B845 (2011) 257-296. DOI: 10.1016/j.nuclphysb.2010.11.019; arXiv:1010.2462; https://arxiv.org/abs/1010.2462
13. J. Lamy-Poirier, P. Mathieu, Path representation of su(2)_k states II: Operator construction of the fermionic character and spin-1/2-RSOS factorization, Nucl. Phys. B847 (2011) 247-273. DOI: 10.1016/j.nuclphysb.2011.01.027; arXiv:1012.4722; https://arxiv.org/abs/1012.4722
14. P. Jacob, P. Mathieu, Particles in RSOS paths, J. Phys. A42 (2009) 122001. DOI: 10.1088/1751-8113/42/12/122001; arXiv:0901.1814; https://arxiv.org/abs/0901.1814
15. G. Feverati, Quantum integrable systems. Quantitative methods in biology, habilitation thesis (2011), arXiv:1101.3722; https://arxiv.org/abs/1101.3722